Biology of HLA class I associated inflammatory diseases

Abstract

Human leukocyte antigen (HLA) class I association is a well-established feature of common and uncommon inflammatory diseases, but it is unknown whether it impacts the pathogenesis of these disorders. The “arthritogenic peptide” hypothesis proposed initially for HLA-B27-associated ankylosing spondylitis (AS) seems the most intuitive to serve as a model for other HLA class I-associated diseases, but evidence supporting it has been scarce. Recent technological advances and the discovery of epistatic relationships between disease-associated HLA class I and endoplasmic reticulum aminopeptidase (ERAP) coding variants have led to the generation of new data and conceptual approaches to the problem requiring its re-examination. Continued success in these endeavors holds promise to resolve a Gordian Knot in human immunobiology. It may ultimately benefit patients by enabling the development of new therapies and precision tools for assessing disease risk and predicting treatment responses.

A –. INTRODUCTION

Associations of the major histocompatibility complex (MHC) locus – termed human leukocyte antigen (HLA) in humans – with immune-mediated inflammatory diseases are well-established. Still, it remains unknown if, and if so, how the HLA contributes to disease pathogenesis. Strong associations exist for HLA class I with inflammatory disorders, many of which are widely known, such as ankylosing spondylitis (AS) and acute anterior uveitis (AAU), psoriasis (PS), and Behçet’s disease (BD). Some are much less so, such as Birdshot chorioretinopathy (BCR), whose association with HLA-A29 is the most powerful known in any human disease. Major HLA I risk loci were first linked to disease more than 50 years ago, but only recent developments in technology and the discovery of epistatic relationships between HLA class I and other constituents of the HLA class I antigen-presentation pathway have provided the tools and the conceptual framework necessary for their fruitful investigation of antigen recognition in the context of HLA. In the following, we try to illustrate some of the recent attempts at disentangling the HLA I challenge and their potential significance for the future of the field.

B –. HLA CLASS I IN INFLAMMATORY DISEASES

Most autoimmune diseases have genetic links to the MHC locus. Typically, strong MHC-disease associations in human immune-mediated diseases exist with MHC (HLA) class I. The classical HLA I molecules (A, B, and C), expressed on the surfaces of most nucleated cell types, play a pivotal role in presenting endogenous peptides to cytotoxic CD8+ T cells and, therefore, have crucial immune-regulatory function. Completely assembled HLA class I molecules consist of a polymorphic heavy chain, an invariant light chain (β2m), and a short peptide, forming a heterotrimeric protein. The MHC is located on the short arm of chromosome 6 (6p21.3–22.1). It exhibits remarkable polymorphism throughout the human population, boasting over 26,341 known unique class I alleles at the time of writing (1). Polymorphic variation in MHC I primarily influences the structure of the binding groove, enabling the presentation of a diverse array of peptide sequences. Genome-Wide Association Studies (GWAS) and other population-based studies have revealed notable associations between polymorphisms in classical HLA class I genes and the susceptibility and/or severity of infectious diseases, diverse autoimmune conditions, and several types of cancer (2). MHC class I molecules bind their peptide cargo within the endoplasmic reticulum (ER) facilitated by the peptide-loading complex (3). Peptides presented by HLA-class I are derived mostly from intracellular proteins and defective ribosomal products (4) that undergo degradation by the cellular 26S proteasome and various proteases in the cytoplasm. A more extensive pool of substrate peptides can be generated during inflammatory or infectious conditions for presentation (5). Subsequently, a minor fraction of the resulting degradation products (6) gains entry into the endoplasmic reticulum (ER) facilitated by the transporter associated with antigen processing (TAP) (7) and are then subject to substrate-specific trimming by the endoplasmic reticulum aminopeptidases (ERAP)1 and ERAP2. Peptides presented on HLA class I tend to be short, typically ranging from 8 to 11 amino acids in length. After many receive a “final cut” by the ER-resident peptidases ERAP1 and ERAP2 they are then loaded onto HLA class I molecules before leaving the ER for the cell surface. There, they engage with antigen-recognition receptors, especially with the T cell receptors of CD8+ lymphocytes. The primary role of this biological system is to facilitate the presentation of self or foreign antigens to T cell receptors (TCR) for immunosurveillance and effector responses, respectively. When aberrations occur within this system, it can trigger an immune response against host cells. Several conditions, including AS, PS, BD, psoriatic arthritis (PsA), and BCR have been linked to disease-predisposing HLA I and aberrant antigen-presentation via their disease-associated HLA class I.

HLA class I association vs HLA I restriction

HLA Class I association

HLA class I association refers to the statistical correlation between the frequency of specific HLA class I alleles and the prevalence of a distinct disease in a defined population. HLA association is, therefore, an epidemiological concept. Noteworthy associations include PS, AS, BCR, and BD with HLA-C*06:02 (8, 9), HLA-B*27 (10), HLA-A*29:02 (11), and HLA-B*51 (12, 13), respectively.

The associations of BCR and AS with HLA I are extremely strong, with almost 100% of BCR patients testing positive for HLA-A*29:02 and 90% of patients with AS for HLA-B*27. In the case of BD, the association with HLA-B*51 is weaker, similar to the one of HLA-C*06:02 with PS, but still substantial and well-established, with odds ratios ranging from 5 to 6.

HLA-restriction

HLA class I restriction is a mechanistic concept of immune function that pertains to the specificity of CD8+ TCR in recognizing and interacting with peptides presented by HLA class I molecules. Each canonical human TCR has a finite number of binding partners restricted to specific HLA molecule-peptide combinations. Restriction refers to the fact that CD8+ T cells are educated to recognize non-immunogenic and immunogenic peptides ligands bound to self HLA class I molecules, and recognition is specific to molecular components of both the peptide and the relevant HLA class I molecule. The former peptides are important for the probing of “self”, maintaining immunosurveillance, the latter may represent intracellular infection or other abnormal cellular activity on the cell surface and can trigger immune responses. Expansion of CD8 T cells, for the most part, indicates recognition of peptide (p)-HLA by the cognate CD8 TCR and triggering of downstream signaling. The immunogenicity (or the lack thereof) can be influenced by specific ERAP1 allotypes that differentially trim peptides which are then presented within a defined HLA I restriction context. The peptide properties conferred by differential trimming may determine whether 1) a given peptide is presented by a specific HLA class I molecule, 2) presented by different HLA class I molecules, or, 3) not presented at all. These properties impact if, towards which antigens and with what preference an HLA-restricted immune response occurs – a phenomenon called immunodominance.

Evidence suggesting HLA class I restriction in inflammatory disease pathogenesis

While HLA I disease associations imply a mechanistic role for the involvement of a specific allele in the pathogenesis of that disease, they do not prove it. As of today, no conclusive evidence has been generated. There are, however, recent studies that strongly suggest a role for HLA class I restriction in the three major human HLA class I associated autoimmune diseases, AS, PS, and BD.

The probably strongest evidence for HLA restriction in a human disease so far was provided by Yang et al., 2022 (14) for HLA-B27-bound peptides recognized by HLA-B27+ AS and AAU-associated TCR. Utilizing TCR-driven selection of HLA-B27:05-based yeast display peptide libraries, they isolated HLA-B27 restricted peptides cognate to disease-associated TCRs from the joints of individuals with HLA-B27+ AS and from the eyes of individuals with HLA-B27+ AAU/AS, and identified conserved peptide hotspots in peptides derived from both microbes and self that were important for T cell recognition.

Another important study offers strong mechanistic clues to the importance of HLA class I restriction in inflammatory disease; in this case, psoriasis: Arakawa et al. 2015 (8) identified a peptide derived from a disintegrin and metalloproteinase with thrombospodin motifs-like protein 5 (ADAMTSL5) as an HLA-C06:02–restricted melanocytic autoantigen targeted by the CD8+ Vα3S1/Vβ13S1 TCR in psoriasis lesions. This TCR specifically attacks melanocytes, the sole epidermal cells expressing ADAMTSL5. Indeed, Arakawa et al., provided evidence of an autoreactive TCR obtained directly from lesion-infiltrating CD8+ T cells recognizing HLA-C06:02-restricted melanocyte-derived peptide in psoriasis, the processing of which was controlled by ERAP1 allotypes (8), as extensively reviewed by Prinz et al. (15).

Cavers et al., (16) published data that support HLA class I restriction in the pathogenesis of BD by showing that reduced activity of ERAP1 in genome-edited HLA-B51+ ERAP1 knock out (KO) antigen-presenting cells changes the immunogenicity to CD8⁺ T cells, influenced by an HLA class I peptidome containing under- trimmed peptides. This models the risk-conferring ERAP1 haplotype 10 (Hap10) which exhibits extremely low trimming activity for most peptides. The authors also found significant shifts in the abundance of antigen-experienced vs naïve CD8 T cells, depending on ERAP1 Hap10 status in an HLA-B51 background in a cohort of untreated, active BD patients and healthy donors (HD). Reconciling these findings conceptually requires HLA restriction and they suggest that ERAP1 Hap10 contributes to Behçet's disease pathogenesis by producing an altered set of HLA-B51-restricted peptides, leading to a shift in the immunodominance of the resulting CD8⁺ T cell response.

ERAP as a modulator of HLA class I mediated immunity

ERAP function in the HLA class I antigen presentation pathway

ERAP1 and ERAP2 are ER-resident enzymes whose main function is to trim peptides that have entered this compartment from the cytosol through TAP, a length-selective peptide transporter, located in the ER membrane, to the length required for stable loading onto HLA class I molecules. The enzyme is called ERAAP in mice where it is the only ER-resident peptide trimming peptidase. While all humans express ERAP1, 75% also express ERAP2 which differs in peptide trimming specificity, although at least some redundancy with ERAP1 likely exists (17). ERAP genes thus serve as an immunomodulator in the HLA class I antigen presentation pathway and some of their polymorphisms in humans have been found to be in epistasis with risk-conferring HLA I. Since ERAP and HLA I are both part of the same pathway, a causative role for that pathway in HLA I-associated disease pathogenesis is likely. The antigen processing pathways are comprehensively reviewed by Blum, Wearsch, and Cresswell (18).

The heterotrimeric HLA I complexes with β2m and intracellular peptides within the ER. These complexes are integral components of a multimolecular peptide-loading complex, including calnexin, calreticulin, tapasin, and the thiol oxidoreductase ERp57. ERAP1 and ERAP2 are encoded by two genes situated on chromosome 5q15 in opposing orientations. These genes exhibit a high degree of polymorphism, with ERAP1 being particularly rich in single nucleotide polymorphisms (SNPs) compared to ERAP2. Several studies have associated variations in these genes with alterations in their functionality. Such modifications have been implicated in the initiation of MHC I-associated disorders, including in infectious diseases (19, 20). ERAP1 exerts considerable influence on the HLA I-bound peptidome, which suggests its potential to generate either immunogenic or tolerogenic immune responses (19, 21-23). ERAP1’s substrate preferences are based on peptide length and the C-terminal residue. Functioning as a unique molecular ruler, ERAP1 is involved in the trimming of peptides from their N-terminal ends to the final lengths (typically 9mer), appropriate for loading into the tight binding groove of HLA I molecules (24). Several studies demonstrated that inhibiting ERAP expression disrupts the p-MHC I repertoire (25-27). Peptides showcased through the 'cytosolic' or 'HLA class I' pathway can either arise from intracellular proteins, be introduced during intracellular infections, or from extracellular antigens cross-presented by dendritic cells. P-HLA complexes subsequently migrate to the cell surface, where they become targets for recognition by matching specific TCR on CD8+ T lymphocytes (Fig. 1). The existence of foreign or modified self-peptides presented by MHC class I molecules enables CD8+ T cells to identify intracellular abnormalities and instigate the elimination of these cells. The efficient generation of the p-MHC I repertoire is also crucial for effective immune surveillance by the CD8+ T cell repertoire (28). Hence, variations in ERAP1 activity, caused by genetic polymorphisms, have the potential to impact the length and biochemical properties of the peptide repertoire presented by HLA I molecules, forming p-HLA I. ERAP thus further diversifies the already extremely diverse human MCH class I system, allowing additional fine-tuning of the specificity of TCR-mediated responses through p-MHC recognition. While this diversity likely represents evolutionary adaptation to selection pressure from infectious pathogens, it may come at the price of “off-target”, i.e., effector immune responses, to self and susceptibility to autoimmune diseases. Indeed, variations in enzymatic activity of ERAP1 have been linked to an increased risk of autoimmune or inflammatory diseases (29). Notably, almost all of the presumably immune-mediated diseases in whom ERAP polymorphisms confer risk are known HLA class I associated disorders, and in most of them, ERAP is in epistasis with the respective risk conferring HLA molecule.

Figure 1.

HLA (MHC) class I antigen presentation pathway. Proteins fragmented by the proteasome in the cytoplasm are translocated into the endoplasmic reticulum (ER) via the transporter associated with antigen processing (TAP). There, the aminopeptidase ERAP1 trims peptides from the N-terminal end for loading onto the major histocompatibility complex class I (MHC I). Peptide-MHC I complexes reach the cell surface through the secretory pathway, where they become visible to the T cell receptors (TCR) on CD8+ T cells. Canonical α/β TCR require specific peptide-MHC combinations for recognition – a phenomenon called MHC restriction.

Epistasis between ERAP haplotypes and disease-associated HLA class I modulates disease risk

Epistasis denotes the influence of one gene on the functionality of another (30). The ERAP1 rs30187 polymorphism increases the susceptibility to AS, but exclusively in conjunction with HLA-B*27 (31). Convincing evidence from several immunogenetic studies also indicates that variant ERAP1 increases the risk associated with certain inflammatory diseases in epistasis with disease-associated HLA I alleles while providing protection against others: rs10050860 and rs17482078 in ERAP1 increase the risk for Behçet’s uveitis/ BD in HLA-B*51 carriers, but also protect from AS (30, 31) and PS (32) in the presence of their disease-associated HLA I alleles, HLA-B*27 and HLA-C*06, respectively. Since these disease risk-modulating epistatic relationships involve two genes in the same pathway, i.e., the intrinsic antigen-presenting pathway, they suggest a disease-related mechanism affecting antigen presentation (32-34). This could be mediated through alterations in substrate preferences of allotypic ERAP1, differences in trimming activity resulting in differential peptide length, the generation of newly immunogenic peptides, the elimination of immunogenic peptides, or other mechanisms (23, 35). Similar considerations may apply to ERAP2, which has disease associations with psoriasis, AS, Crohn’s disease, and BCR (11, 36-38).

Haplotypes of ERAP1 are identified as risk factors among HLA-B*51-positive individuals in BD, while no known association exists with ERAP2 for this disease. Several ERAP1 SNPs have been recognized for their influence on the enzyme's function (39). The proposed mechanisms accounting for this effect involve direct interactions with the substrate (40), the level of protein expression (41, 42), effects on conformational dynamics (43), or a combination of these factors (38). These functional differences may provide a stepping stone for disentangling immune mechanisms mediating the biological effects of the observed epistasis between ERAP1 and HLA (44, 45).

ERAP1’s impact on HLA class I-restricted immune responses

The evolving knowledge about ERAP haplotypes and the varying trimming activities linked to its respective allotypes, invariably raises the question of if and how those properties impact HLA class I-restricted immunity.

Arguably the most compelling evidence for ERAP1-dependent HLA-B27 restriction comes from a study wherein ERAP1 with low trimming function associated with failure to generate hepatitis C immunodominant 9-mer epitopes for presentation on HLA-B27 – typically protective against hepatitis C virus (HCV). Instead, there was an alteration of immunodominance toward 10 and 11-mer non-HCV reactive peptides, ultimately leading to impaired viral clearance of HCV (46).

For presumably immune-mediated diseases, the currently most compelling experimental evidence for ERAP1-modulated HLA class I restriction was provided through the above-mentioned studies by the Prinz group. The low-trimming (HLA-B*51 BD risk-associated ERAP1 allotype) Hap10 resulted in less efficient generation of the autoantigenic (HLA-C*06:02 PS-associated) epitope (ADAMTSL5-derived peptide) compared to PS risk haplotypes of ERAP1. On the contrary, an ERAP1 risk haplotype for PS led to more efficient production of the autoantigen, increased HLA-C expression and significantly enhanced stimulation of the psoriatic TCR by melanocytes compared to a protective haplotype. These findings are conceptually important as they strongly suggest, in the context of an HLA class I associated disease, here, psoriasis, the modulation of a disease risk-associated HLA class I restricted immune response through ERAP1-dependent autoantigen processing by an ERAP1 risk haplotype.

Lastly, evidence for the significance of ERAP1 in the modulation of HLA-B51 mediated immunity is evolving through the demonstration of the impact of absent or low ERAP1 activity on the immunogenicity of allogeneic CD8 T cells (16), which is linked to length differences in the differentially ERAP-trimmed HLA class I peptidomes and reflected in phenotypic shifts between antigen-experienced CD8 T cell populations in HLA-B51 carrying humans with and without BD, depending on their ERAP1 status.

Direct evidence for ERAP1 modulation of autoantigen-specific T cell responses in HLA-B27-associated AAU and AS, as well as in HLA-B51 associated BD, which would require the identification of HLA-B51 restricted peptides cognate to disease-associated TCR and their subsequent ERAP1 dependent modulation, is still missing.

Beyond indirect indications of ERAP1-mediated modulation of immunodominance, a substantial and replicated body of research in mice offers compelling evidence. This evidence stems from viral infections, including lymphocytic choriomeningitis virus (LCMV), murine cytomegaly virus (mCMV), and vaccinia virus, where ERAP1 KO significantly disrupted the hierarchy of CD8+ responses to viral epitopes, resulting in a shift in immunodominance and the incapacity to mount an effective anti-viral immune response (47, 48).

Interestingly, recent investigations in HLA-B27 transgenic ERAP1 deficient rats have suggested a mechanism unrelated to HLA class I restriction as an explanation of the epistatic interplay between ERAP1 and HLA-B27. Complete loss of ERAP1 provided partial protection against HLA-B27-induced peripheral arthritis and epididymoorchitis (49) and was related to reduced misfolding of HLA-B27 and reduced ER stress. These findings align with previous studies suggesting that factors such as HLA misfolding, intracellular dimerization, and the slow assembly phenotype rather than HLA allotype-specific peptide binding partake in disease causation (50, 51).

C –. UNRESOLVED QUESTIONS AND AREAS OF UNCERTAINTY

Target peptides and their physiological relevance

The technical ability to identify HLA-restricted peptides cognate to disease-associated TCR in an unbiased fashion is an enormous step forward in the field and has yielded the first exciting results. Yang et al., demonstrated that a bacterial conserved peptide, YEIH, derived from microorganisms associated with reactive arthritis ReA, is recognized by disease-associated TCR in HLA-B27 restriction (14). This peptide exhibits conservation across various bacterial species, including Escherichia coli, Salmonella, Shigella, and Klebsiella. Yang et al., have successfully identified several other peptides derived from human or microbial proteins that are recognized by disease-associated TCR in HLA-B27 restriction. The recognition of these peptides by disease-associated TCR suggests their (or that of structurally similar peptides) partaking in disease pathogenesis. Direct disease relevance of these peptides in vivo, in the context of AS or AAU, still remains somewhat uncertain, however. Indeed, there currently is no published evidence of the occurrence of these peptides in the peptidomes of elution studies conducted on HLA-B27-transfected cell lines or using cells from HLA-B27 transgenic rats – none of which are likely to accurately represent the disease – and studies on human tissue are still missing at the moment.

Of similar importance to the field is the above-cited Arakawa et al’s work in PS (8). Here, the authors could identify a cell target, i.e., melanocytes and were able to show the capacity of the disease- associated antigen to cause IL-17 secretion in PS patient derived peripheral blood mononuclear cells (PBMC)/ CD8 T cells.

Caver et al’s work (16) in BD demonstrates the ERAP1-dependent length distribution of pan-HLA I-derived peptides and their HLA-B51 deconvoluted fractions. The authors performed microbiological homology analyses of HLA-B51- binding, long (10, 11-mer) peptides experimentally identified in the ERAP1 KO (mimicking the BD risk variant of HLA-B51/ERAP1 Hap10) conditions of their experimental system. They found more than half of the HLA-B51 restricted peptides in that group (63%) had non-negligible foreignness scores (52) suggesting some degree of homology with experimentally determined microbial-derived epitopes, making TCR cross-reactivity resulting in immunogenicity likely.

Despite the recent progress in all of the three major HLA class I-associated rheumatic disease entities, AS/AAU, PS, and BD, there remains a paucity of data linking “disease-associated” peptides in their disease-relevant HLA class I restriction directly to active human disease. It also remains unclear at the moment whether tissue specificity is an important factor, whether modification of self-peptides in the cytosolic pathway is the primary source, or whether microbial peptides introduced through intracellular infection or cross-presentation in DC, molecular mimicry, or other factors play a major role.

Peptide origin and recognition

Molecular mimicry is an immunological phenomenon marked by cross-reactivity of antigen-recognition receptors between host and non-host-derived, often bacterial, antigens, given their structural similarity. Molecular mimicry includes the recognition of such antigens in HLA I restriction contexts by TCR. This can occur by presenting peptides derived from self-proteins or homologous to self-antigens but originating from microbial proteins. Much of the recent work, in particular Yang et al’s studies (53) seems to revive the old concept of molecular mimicry. Still, the concept of molecular mimicry's mechanisms poses numerous challenges to fully comprehend the intricate interplay among host genetic factors, environmental triggers, and the immune system (54), including prolonged periods of latency prior to the appearance of disease.

A common ‘suspect’ supplier of proposedly pathogenic peptides is the microbiome. There is growing evidence linking the microbiome to a wide range of rheumatic diseases in general (55). Most studies in human microbiome are associative, indicating a correlation between imbalances in gut microbiota (dysbiosis) and inflammatory or autoimmune conditions (56). Recent research has revealed the impact of the gut microbiome on various diseases associated with HLA genes. Phoebe et al. (57) demonstrated that HLA-B*27 status correlates with modifications in cecal microbiota. By employing biome representational in situ karyotyping (BRISK) and 16S rRNA gene sequencing, the team identified variations in the cecal microbiota of transgenic Lewis rats carrying HLA-B*27 and human β2-microglobulin (hβ2m) compared to wild-type Lewis rats. Notably, the analysis revealed species-specific differences, such as an increased abundance of Bacteroides vulgatus in HLA-B*27/hβ2m and hβ2m rats compared to the wild type.

In a mouse model of spontaneous uveitis, Caspi and colleagues demonstrated that retinal antigen- specific T cells were activated by cross-reactivity with a commensal microbial antigen in the gut, triggering autoimmune reactivity in the eye (58), providing compelling experimental evidence that the commensal microbiome may be causally linked to autoimmunity in mice.

HLA I and other MHC pathway constituents as potential risk stratifiers and therapeutic targets

While simple risk stratification by HLA class I/ ERAP carrier status is possible through the extrapolation of data from immunogenetic studies, it is not feasible in clinical care because the overall contribution of HLA to disease risk is not large enough to justify genetic testing without clinical suspicion in most cases. It is also not universally applicable as both HLA and ERAP haplotype allele frequencies are ethnicity-dependent. The combination of such profiles with other genetic, epigenetic or transcribed markers, however, might lead to a more comprehensive assessment of lifetime disease risk, or even to diagnostic tests for these complex-polygenic diseases in the future.

More realistic in the short run, is the application of HLA and ERAP profiles in assessing treatment responses to therapies that will be targeted to the MHC I pathway itself, such as the use of ERAP1-specific activity enhancers or inhibitors that could be rationally applied to patients with specific HLA/ ERAP genotypes.

Even in the absence of precise mechanistic understanding, risk stratification by HLA/ERAP might be useful: it is dispensable to know which peptides are presented through the action of ERAP with a specific enzymatic activity, if modulating ERAP1 activity confers a therapeutic/ protective effect.

Another strategy involves manipulating potentially relevant antigen supply, such as through manipulation of the microbiome (59). The microbiome plays a significant role in immune regulation, and strategies to promote its restoration might help rebalance immune responses and alleviate MHC I-associated diseases. The emerging approaches for T cell antigen discovery present a promising avenue for future research in this context (60). In instances of autoantigen-mediated pathology, T cell tolerization or expansion of autoantigen-specific regulatory T cells are other potential approaches that could help dampen auto-immune pathology.

The idea that HLA class I-restricted autoimmune responses are probably targeting a particular cell expressing the parental protein of the immunogenic self-peptide can be pivotal in identifying these peptides. This concept can facilitate the determination of the HLA restriction of TCRs and narrows down potential T cell epitopes to the transcriptome or immunopeptidome of the target cell. Utilizing peptide libraries to determine the recognition motif of pathogenic TCRs enables the exploration of proteins containing homologous peptide sequences (61).

The field of therapeutic modulation of ERAP1 activity is still in its early stages, marked by the exploration of pharmacologic inhibitors and enhancers which have been characterized in terms of selectivity, potency, and their effect on antigen presentation in vitro (54, 59, 60, 62-66) but their activity in vivo has not yet been comprehensively characterized (62, 64, 67-69). Ultimately, the success of these compounds hinges on their targeted application to genetically defined patient groups and subsets. Findings indicating that compounds inhibiting ERAP1 activity in one allotype may enhance activity in others further complicate those efforts (62).

Non-canonical pathways

KIR and LIRL in HLA-I associated inflammatory diseases

Antigen-recognition receptor recognition of HLA class I is not limited to p-HLA-TCR interactions. Killer cell immunoglobulin-like receptors (KIR) and Leukocyte Ig-like receptor (LIRL) are important receptors on natural killer (NK)/ natural killer T (NKT) and dendritic cells (DC), respectively, which can recognize HLA I and whose recognition of the latter can be influenced by peptide-HLA binding without involving HLA restriction per se.

KIR’s HLA class I ligands have been investigated in diverse diseases, encompassing autoimmune disorders, infectious diseases, and cancers (70, 71). NK cells engage with HLA class I molecules through two distinct categories of receptors: KIRs and inhibitory CD94:NKG2 receptor. KIRs constitute a family of receptors with both inhibitory and activating functions, primarily expressed on the surface of NK cells and, to a lesser extent, on some T cells.

The field has been strongly influenced for decades by the notion that cells expressing HLA-B27 may up-regulate the unfolded protein response (UPR) in response to misfolding and protein aggregation, facilitated by a tendency of HLA-B27 to form homodimers (72) that may be recognized by (KIR)3DL1 receptors on NK cells (73). The formation of homodimers could be facilitated by – or result in – the loading of unstable and suboptimal peptides with a rapid dissociation rate. This might implicate ERAP1 variants in the generation of such peptides, ultimately leading to the activation or inhibition of KIR receptors through HLA-B*27 ligands. Both KIR3DL1 and KIR3DL2 recognize cell surface homodimers, and the levels of KIR3DL2 have been found to be significantly up-regulated in HLA-B*27 positive AS patients (74). Saunders et al. demonstrated that KIR3DL1 recognition of HLA class I ligands is influenced by both the peptide sequence and conformation, as dictated by the HLA polymorphic framework (75). An additional finding indicated that ERAP1 could potentially impact NK cell responses by modulating inhibitory receptors such as NKG2A/CD94, which are also expressed by CD8+ T cells, targeting the non-classical HLA-I molecule HLA-E (76). Due to the involvement of the signal sequence derived from HLA-I molecules, including those in HLA-A*29, HLA-B*27, and HLA-B*51, in HLA-E's inhibitory function, it follows that ERAPs might impact NK cells and CD8+ T cells through the trimming of MHC I molecule-derived peptides, as demonstrated in cancer models (77). While KIR receptors can detect alterations in the immunopeptidome induced by ERAP1, it is noteworthy that KIR genes do not appear to impact the ankylosing spondylitis risk associated with HLA-B*27 and ERAP1 (78, 79). This observation might imply that the mechanisms underlying disease mediated by ERAP1 and HLA-I may have less reliance on KIRs. In the context of BCR, the involvement of NK cells and KIR receptors in the disease's pathogenesis cannot be dismissed. This is particularly relevant due to potential ERAP2-specific modifications on the peptidome of HLA-A*29:02, which may impact the interaction between HLA and NK KIR receptors (80, 81).

Besides, HLA class I molecules serve as ligands for the leukocyte immunoglobulin-like receptors (LILR), with LILRB1 and LILRB2 being the most extensively characterized among them. The interaction of HLA class I molecules with LILR has been linked to disease susceptibility, as evidenced by the contribution of this interaction to psoriasis susceptibility (82). LILRB1 and LILRB2 are inhibitory receptors primarily expressed on myeloid cells, including dendritic cells and macrophages. Signaling through LILR influences the activation of these antigen-presenting cells (83), but little is known about their interaction with specific HLA class I molecules.

B27 homodimer/ER stress

The conventional heterotrimeric MHC class I molecule consists of three individual polypeptides that are not covalently linked. These include a highly variable heavy chain, a β2-microglobulin (β2m) light chain, and an oligopeptide, usually comprising 8 to 10 residues in length (84, 85). Following the synthesis and glycosylation of MHC molecules in the ER, chaperones such as calreticulin and tapasin initially stabilize the free heavy chains until they attain a conformation suitable for binding β2m and a peptide of the appropriate length. Subsequently, emerging MHC class I molecules commonly engage with antigenic peptides and convey them to the cell surface, facilitating presentation to the TCRs on T lymphocytes.

Formation of B27 homodimers (B27₂) and ER stress have been implicated in the pathogenesis of inflammatory diseases, particularly spondyloarthritis (86). The capacity of HLA-B27 to generate β2m-free homodimers through Cys67-mediated disulfide bonding (referred to as B27₂) was initially observed during the in vitro refolding of recombinant HLA-B27 (87). HLA-B27 has been shown to misfold during assembly, leading to ER stress and autophagy responses, which may contribute to the dysregulation of immune responses and the development of autoimmunity (88). Misfolding of HLA-B27 in the ER and the activation of the UPR is thought to lead to the production of interleukin-23 and other effector immune responses (88). Several forms of the B27 free heavy chain, such as B27₂, are expressed on the cell surface through endosomal recycling of heterotrimers, constituting a non-conventional pathway (89). Due to misfolding and ER stress, cell surface HLA-B27 homodimers (B27₂) and free heavy chains interact with innate immune receptors on T, NK, and myeloid cells. These receptors include human receptors such as KIR3DL1, KIR3DL2, and LILIRB2 (90, 91), as well as rodent receptors known as paired immunoglobulin receptors (PIR). These interactions with immune receptors may contribute to the pathogenesis of spondyloarthropathies (SpA), potentially through the recognition of non-canonical conformations of HLA-B27 by these receptors. The propositions suggesting the involvement of cell surface B27₂ in the pathogenesis of AS/ SpA have been made based on certain observations. However, there remains limited direct evidence in human SpA to date.

D –. CONCLUSIONS

HLA class I-associated inflammatory diseases are among the most common non-infectious inflammatory disorders worldwide. Their hallmark feature, the HLA class I association, was discovered more than 50 years ago for some of them. Yet, a mechanistic understanding of the potential role of their associated HLA alleles in disease causation is missing to date. Despite the notion of pathogenic p-HLA activating disease-driving TCR remaining the most intuitive hypothesis to date, older evidence from rat models and the difficulties in identifying disease antigens has kept the “arthritogenic peptide hypothesis” in the shadow of other potential explanations for decades, such as HLA-B27 triggering an ER stress response.

Recent data have challenged this view, facilitated by technical advances allowing the reconstruction of the alpha and beta chains of the human TCR from disease effector sites through novel sequencing techniques and the ability to interrogate those receptors in a desired HLA restriction context in a largely unbiased fashion. Promising candidate peptides have now been identified for HLA-B27-associated AS and AAU, PS, and there is experimental evidence suggesting HLA-restriction in subsets of BD, rendering target peptide identification in the future very likely here, too. The actual disease relevance of these peptides remains to be fully established, which is the goal of ongoing and future studies. In the same vein, understanding the molecular effects of ERAP1/2, which trim peptides for loading onto HLA class I and whose coding genes are in epistasis with disease-associated HLA, remain essential research areas. Understanding their contribution to disease pathogenesis may help explain risk conferring and protective effects, allow the development of ERAP activity-modulating therapy, its rational testing in clinical trials, and, ultimately, the prediction of treatment responses based on genotype, beyond disease phenotype alone.

PRACTICE POINTS.

• HLA class I-associated diseases are centrally important disorders in rheumatologic practice

• Results of HLA class I typing are of limited use for the diagnosis of associated diseases and have no direct impact on clinical care at this point

• Combining HLA typing with additional genetic and other information may facilitate the diagnosis, risk-stratification, and prediction of treatment responses in HLA I-associated diseases in the future

RESEARCH AGENDA.

• The continued identification of HLA class I-restricted peptides and disease-associated TCRs remains an important research target in HLA I-associated diseases

• Currently known candidate peptides in HLA-B27-associated disease are an important step forward and require further validation to assess their biological disease relevance

• Assessment of non-HLA constituents within the HLA class I antigen presenting pathway on p-HLA repertoires and their physiological consequences are promising research targets

Abbreviations

ADAMTSL5

a disintegrin and metalloproteinase with thrombospodin motifs-like protein 5

AAU

acute anterior uveitis

AS

ankylosing spondylitis

BCR

birdshot chorioretinopathy

BD

Behçet’s disease

BRISK

biome representational in situ karyotyping

β2m

β2-microglobulin

DC

dendritic cell

ER

endoplasmic reticulum

ERAP

endoplasmic reticulum aminopeptidase

Hap10

haplotype 10

HD

healthy donor

hβ2m

human β2-microglobulin

HLA

human leukocyte antigen

KIR

killer-cell immunoglobulin-like receptors

KO

Knock out

LILR

leukocyte immunoglobulin-like receptor

LCMV

lymphocytic choriomeningitis virus

MHC

major histocompatibility complex

mCMV

murine cytomegaly virus

NK

Natural killer

NKT

Natural killer T

PIR

paired immunoglobulin receptors

p

peptide

PBMC

peripheral blood mononuclear cells

PS

psoriasis

PsA

psoriatic arthritis

SNP

single nucleotide polymorphism

SpA

spondyloarthropathies

TCR

T cell receptor(s)

TAP

transporter associated with antigen processing

UPR

unfolded protein response